5 generative AI takeaways for CEOs

1. It’s not time to “test the waters.” It’s time to scale

A lot has happened since late last year when generative AI models, which use deep learning to create images, text and other kinds of content from prompts, entered the mainstream. It’s now already possible to operationalize generative AI at scale: adapt existing models to deploy them in multiple functions and lines of business. Real-world examples business executives provided included enhanced customer service, automated fraud alerts, optimized workforce scheduling and intelligent automation to enable healthcare workers to do more with less. At PwC, with the help of a $1 billion investment, we’ve built an “AI factory” for ourselves and our clients to help scale generative AI responsibly and quickly. 

2. Generative AI will change everyone’s jobs

The democratization of tech isn’t new. Your company may already be using “low-code” and “no-code” tools. But public domain and enterprise-class generative AI models can take democratization to a whole new level. Almost anyone in your company who works with knowledge, even if they have no technical background, may soon be working directly with generative AI. But they’ll need new skills to use generative AI well, so that it can deliver relevant and consistent outputs. You’ll also need talent in new specialized roles, such as “prompt engineer” and “model mechanic,” to help adapt generative AI models to your unique needs and to establish appropriate practices that can help drive towards quality outcomes. As part of a “generative AI factory,” these specialists will also need to work closely with strategy and product leads, and specialists from risk, legal (think privacy, IP protection, copyright), data and even procurement (think licensing). 

3. Responsible AI is both urgent and a competitive differentiator

Generative AI doesn’t just offer new ways to turbocharge productivity and enable new business models. It also offers four broad sets of risks, related to data, bias, prompts or inputs, and human-tech interaction. Fortunately, the same responsible AI methodology that’s designed to enable conventional AI can also help establish a framework for good governance over the use of generative AI. Given the scale and speed at which generative AI is developing, it’s critical to deploy trust-by-design: embed governance, risk management, and controls designed to help address human-tech interaction in generative AI from the start. Responsible practices in deploying generative AI can not only manage risks, they can also help your stakeholders trust the outputs that your AI produces, and trust that your organization is using it ethically. This can potentially offer you a competitive advantage in both the market and the competition for talent.

4. What you build and how you innovate will change

We’ll likely see a new kind of build-versus-buy debate take shape. In the past, that was usually reserved for enterprise apps — do you need a bespoke application or will off-the-shelf do? That now applies to different kinds of R&D, especially new products and services. As generative AI models continue to advance, they may quickly outpace the innovations that companies have been investing in for years. Some of the boldest may be willing to cut their (significant) losses and start over with these models at the core, focusing on the ten percent that’s differentiating for their market or industry. Instead of developing a product or capability from scratch, they may concentrate on how to fine-tune the foundational AI model for their specific applications. They may also take a close look at key assets as they formulate their strategy, including the data they have (or could synthetically generate), the industry and functional experience they have, and their ecosystem business partners. The companies and technologies that they choose to build the future with, will likely be critical relationships.

5. Culture may determine who wins and loses

With everyone in your organization a potential generative AI user, your organizational culture will be more important than ever: Your people should be able to embrace experimentation, iteration and persistence. Generative AI is already poised to transform how knowledge workers get their jobs done. Domain specialists will have to learn to work with generative AI. Many technologists (including software engineers) will likely need to transition to become generative AI specialists. Data scientists who formerly did nearly everything AI related should focus on more specific roles. And generative AI is still advancing. New and improved foundation models (the “brain” of generative AI) are being released. New services and tools that work on top of these models are also constantly emerging. If your people aren’t willing to learn new skills and (when needed) fail fast and move on, your organization may not be able to pivot to capture new opportunities.